jdupes vs rdfind

jdupes is a free utility for finding and handling exact duplicate files on Linux, macOS, and Windows. It identifies files that are fully identical rather than matching images, sounds, or other formats by similarity. Matching can use file size, hashes, and byte-for-byte comparison, with optional partial-hash matching. The project says it checks matching data after hashing to reduce the chance of false matches from hash collisions. Found duplicates can be reviewed before deletion, deleted without prompting, linked, summarized, or handled through copy-on-write deduplication. On Linux and Unix, the instructions use command-line directory paths; the project also provides a native Windows port, including support for Unicode. Docker use is documented, and standard output can feed shell scripts. A database can retain scan metadata for repeat runs, though the project labels that feature as under development and notes quirks. The README warns that quick matching skips byte-for-byte checks and can classify non-duplicates as duplicates. It also warns that disabling file-change checks creates a security vulnerability, especially when linking or deleting. jdupes is distributed under the MIT License.

rdfind is a free command-line utility that finds duplicate files by comparing contents rather than names. It searches supplied directories recursively, including comparisons within or across directories, and can also accept individual files. Matching uses file size, first and last bytes, checksums, and inode or device identity. Supported checksums are MD5, SHA-1, SHA-256, SHA-512, and xxh128, with SHA-1 as the default. By default, rdfind writes a results file without changing files. Options can delete duplicates or replace them with symbolic or hard links, while a dry run previews an enabled action. Filters can ignore empty files or set minimum and maximum sizes. It runs on Linux, macOS, and Windows; Linux packages include apt and dnf examples, Mac installation can use MacPorts, and Windows users may use Cygwin. The manual warns about changing directories during a run, symlink attacks when deleting or linking, and unsupported ownership or permissions handling.